Evaluation of ANN-BP and ANN-GA Models Performance in Predicting Mechanical Properties and Machinability of Cast Copper Alloys

In this paper artificial neural network (ANN) models were developed to predict the mechanical properties and machinability of Cu–Sn–Pb–Si–Ni– Fe–Zn–Al alloys on the basis of the chemical composition (wt%) of alloying elements. The multi-layer perceptron architecture was used for developing ANN models. Two ANN training approaches, namely, the classical gradient descent back propagation (BP) and genetic algorithm (GA), were applied and statistically compared. The statistical methods of root mean square error (RMSE), absolute fraction of variance (r) and mean absolute percent error (MAPE) were used for evaluating the performance of the developed ANN models. The results showed that training with GA improved the prediction performance of ANN models. By taking the full potential of GA through fine tuning of the GA parameters, the effectiveness of the approach could be further improved allowing for a wide application in the area of material engineering for the prediction of mechanical properties.